Balanced transcription of cell division genes in Bacillus subtilis as revealed by single cell...

Balanced transcription of cell division genes inBacillus subtilis as revealed by single cell analysis

Erik Nico Trip,1† Jan-Willem Veening,2*†

Eric J. Stewart,3 Jeff Errington4 andDirk-Jan Scheffers1**1Department of Molecular Microbiology and 2MolecularGenetics Group, Groningen Biomolecular Sciences andBiotechnology Institute, Centre for Synthetic Biology,University of Groningen, Nijenborgh 7, 9747 AGGroningen, the Netherlands.3Department of Biology, Northeastern University,Boston, MA, USA.4Centre for Bacterial Cell Biology, Newcastle University,Newcastle upon Tyne, UK.

Summary

Cell division in bacteria is carried out by a set ofconserved proteins that all have to function at thecorrect place and time. A cell cycle-dependent tran-scriptional programme drives cell division in bacteriasuch as Caulobacter crescentus. Whether such a pro-gramme exists in the Gram-positive model organismBacillus subtilis is unknown. Here, we investigate therole of gene transcription as a potential regulatorymechanism for control of division in B. subtilis. Tran-scriptional GFP fusions in combination with flowcytometry demonstrated a constitutive promoteractivity, independent of growth rate, of nine tested celldivision genes. These measurements were verified byquantitative real-time reverse-transcription PCR(qrtPCR). Time-lapse fluorescence microscopy wasperformed on a set of selected reporter strains to testtranscriptional regulation during the cell cycle. Inter-estingly, although the average fluorescence remainedconstant during cell-cycle progression, individualcells demonstrated a roughly twofold higher promoteractivity at the end of the cell cycle. This cell cycle-dependent increased promoter activity can be partlyexplained by the doubled promoter copy number afterDNA replication. Our results indicate that the tran-scriptional activity of promoters for cell division genes

remains constant regardless of growth rate and cell-cycle state, suggesting that regulation of cell divisionin B. subtilis predominantly takes place at the post-translational level.

Introduction

Cell division in bacteria is a dynamic event during whichone cell divides into two. In Gram-positive bacteria, aphysical barrier, the septum, is constructed at mid-cellacross the diameter of the cell, to divide it in two compart-ments. Subsequently, the septum splits, thereby freeingthe two cells. During division, different processes have totake place in a specific order; elongation of the cell, estab-lishment of a division site at mid-cell, further elongation,constriction and septum formation, and finally separation.These different stages of division require the activity ofmany cell division proteins (CDPs). These CDPs form largeprotein complexes at mid-cell that drive cell division, col-lectively known as the divisome (Adams and Errington,2009). A hierarchical two-step assembly of the divisomehas been reported for the Gram-positive model organismBacillus subtilis. During early division, FtsZ assembles intothe Z-ring at mid-cell assisted by FtsA, ZapA and EzrA. Ata later stage, a second group of CDPs including GpsB,FtsL, DivIB, FtsW, PBP2B (pbpB), PBP1 (ponA) andDivIVA localize at mid-cell to complete the divisome(Gamba et al., 2009). Several of these second stage pro-teins, FtsL, DivIB, DivIC and PBP2B, are recruited coop-eratively to the divisome; one missing CDP in this secondstage of divisome assembly destabilizes interactionsbetween all the other CDPs, stalling division (Erringtonet al., 2003). So, in order for division to proceed correctly,the CDPs have to become active at the correct place andtime.

The positioning of Z-ring formation at exactly mid-cell isensured by the combined effort of the Min-system,nucleoid occlusion and a potential third, as yet unknown,mechanism (Wu and Errington, 2004; Dajkovic et al.,2008; Rodrigues and Harry, 2012). The Min-system is acell pole associated inhibitor of FstZ polymerizationthat prevents initiation of a second round of division at anewly formed cell pole (Gregory et al., 2008). In nucleoidocclusion, the non-specific DNA-binding protein Nocassociates with the replicating nucleoid and inhibitsFtsZ polymerization (Wu and Errington, 2004). During

Received 25 February, 2013; revised 18 April, 2013; accepted 22April, 2013. For correspondence. *E-mail [email protected]; Tel.(+31) (0)50 363 2408; Fax (+31) (0)50 363 2348 or **[email protected]; Tel. (+31) (0)50 363 2319; Fax (+31) (0)50 3632154. †These authors contributed equally.

bs_bs_banner

Environmental Microbiology (2013) doi:10.1111/1462-2920.12148

© 2013 John Wiley & Sons Ltd and Society for Applied Microbiology

mailto:[email protected]

mailto:[email protected]

segregation of the sister nucleoids to the two cell halves,the concentration of Noc at mid-cell is reduced, thusleaving space for Z-ring assembly (Hariharan et al., 1982;Moriya et al., 2010). The balance between polymerizedand non-polymerized FtsZ seems to be maintainedthrough ClpX, but in B. subtilis this seems to be throughdestabilization of FtsZ polymers without degradation ofFtsZ (Weart et al., 2005), whereas in Escherichia coli themechanism is through proteolysis of polymeric FtsZ byClpXP (Camberg et al., 2009).

Another important factor that controls Z-ring formationis nutrient availability. Bacillus subtilis cells maintain aconstant ratio of cell length per Z-ring (L/R) regardless ofgrowth rate; the proportion of cells with Z-rings variesinversely with doubling time, ranging from 85% in LB to50% in minimal sorbitol medium at mid-exponentialgrowth phase (Weart et al., 2007; Chien et al., 2012). TheL/R ratio remains constant because cells are significantlylonger in rich medium compared with poor medium, whilethe diameter of the cells remains constant. The metabolicstate of the cell provides a cue for Z-ring formation andsubsequent division. Bacillus subtilis does this via theglycolipid biosynthetic pathway; PgcA and GtaB transmitcarbon availability to the effector UgtP that inhibits Z-ringformation until the cell reaches a critical mass for divisionto commence (Weart et al., 2007). The cell lengthincreases with growth rate to facilitate multi-fork DNAreplication, in order to speed up the cell-replicationprocess (Sargent, 1975; Sharpe et al., 1998).

Even before Z-ring formation and other division pro-cesses can take place, critical amounts of CDPs have tobe present in the cell. FtsZ can only assemble into a ringand facilitate divisome formation when present above acertain concentration. In steady-state conditions, alreadya small reduction in FtsZ expression leads to a transientdelay in division in E. coli (Rueda et al., 2003; Weart andLevin, 2003). For E. coli, it was found that ftsZ expressionoscillates during the cell cycle with a peak at the onset ofchromosomal replication (Garrido et al., 1993). The peakin expression of ftsZ before the start of division stronglysuggests regulation of the start of division by transcription.However, the expression of genes encoding other (latestage) CDPs during the cell cycle was not studied. ForB. subtilis, oscillating ftsZ expression during the cell cyclehas never been described, but it was found that ftsZexpression is three to fourfold higher when this bacteriumenters the developmental process of sporulation in whichthe cell divides asymmetrically to produce an endosporeat one end of the cell pole. This burst in expression is notcritical for sporulation, although it does increase sporula-tion efficiency (Gonzy-Treboul et al., 1992). Despite theapparent transcriptional cell-cycle regulation in E. coli,other work describes constant protein levels of FtsZ at arange of growth rates in both E. coli and B. subtilis and

fluctuating protein levels of FtsZ during the cell cyclecould not be found for these two model organisms (Weartand Levin, 2003). On a population level, analysis oftranscriptomes by RNA tiling has shown a balancedexpression of cell division genes in B. subtilis when cul-tured in a wide range of growth media and when exposedto stress such as temperature shifts, ethanol and variousdrugs. Cell division genes were shown to belong to theSigA promoter cluster that displays low variance in activ-ity, ensuring stable transcription of genes that encodeproteins with essential functions (Nicolas et al., 2012).Still, the response of transcription of cell division genes todifferent growth rates (e.g. pace of division) was notstudied.

Cell-cycle regulation of FtsZ protein levels as well asftsZ transcription does occur in the differentiating bacte-rium Caulobacter crescentus. During division, a maturestalked cell releases a motile progeny swarmer cell that istemporarily inhibited for replication. Swarmer cells inherithigh levels of the cell-cycle controller protein CtrA thatinhibits ftsZ transcription, while the levels of inherited FtsZprotein are decreased due to proteolysis (Quardokuset al., 1996; Kelly et al., 1998). After the inherited levels ofCrtA decrease, the transcription of ftsZ commences andthe swarmer cell matures into a stalked cell that is able todivide.

Because of the observed differences in transcriptionalregulation of CDPs in various bacteria, we were interestedto study the transcriptional response of cell division genesto various growth rates in B. subtilis. We also wonderedhow transcription of cell division genes is organizedduring the cell cycle. Can transcription be the underlyingprocess that speeds up or slows down division and cantranscription regulate cell-cycle progression? To test this,we created transcriptional fusions to the promoter regionsof nine cell division genes (comprising both early and lateCDP) and monitored their transcriptional response tochanges in growth rate by flow cytometry. In a complimen-tary approach, the transcriptional response of cell divisiongenes to growth rate was analysed at the mRNA level byrtPCR. The expression of cell division genes was alsofollowed during the cell cycle using automated fluores-cence time-lapse microscopy. Our findings demonstratethat, in general, cell division genes in B. subtilis are stablytranscribed independent of growth rate and cell cycle.

Results and discussion

Experimental strategy

The aim of this work was to study the transcription ofgenes coding for CDPs, both at the population and singlecell level at different growth rates, and at the single celllevel during the cell cycle. To do so, we constructed a setof promoter-region fusions to gfp (Fig. 1, Table 1). These

2 E. N. Trip et al.

© 2013 John Wiley & Sons Ltd and Society for Applied Microbiology, Environmental Microbiology

promoter–gfp fusions were integrated at the amyE locus,which provides a stable genetic environment that allowsgrowth of the strains for analysis in the absence of anti-biotics without the risk of loss of the fusion through recom-bination, which is especially important during single cellanalysis by time-lapse microscopy. As a positive control afusion to the rrnB promoter was included, as transcriptionfrom this promoter is strongly correlated to growth rate,while the constitutive veg promoter served as a control forunregulated expression (Krasny and Gourse, 2004). Tomake sure that the read-out from amyE reflected correctpromoter activity, we also generated a subset of promoterfusions at the original locus with the pBaSysBioII tran-scriptional fusion integration vector (Botella et al., 2010),again including the rrnB promoter as a control (Fig. 1).Populations growing at different rates were sampled atmid-exponential phase for each growth medium, when thegreatest percentage of the culture undergoes division,and analysed by flow cytometry to determine promoteractivity. Bacillus subtilis 168 1A700 was used as a hoststrain because this strain grows mainly as motile singlecells, not chains, and therefore can be used for flowcytometry (Veening et al., 2006). To verify the flowcytometry results by an independent technique, the tran-scription of genes coding for CDPs was also monitored

at different growth rates by quantitative real-time reverse-transcription PCR (qrtPCR). For a valid comparisonbetween populations cultured at different growth rates indifferent media the moment of sampling is critical; allcultures have to be sampled at the mid-exponentialgrowth-phase. For this matching, we determined the spe-cific mid-exponential growth-phases for cultures grown inall used media, and sampled at those moments. Finally,promoter activity in the amyE promoter–gfp fusion strainswas monitored using time-lapse microscopy on growingmicrocolonies, which allowed the study of promoter activ-ity in single cells during the cell cycle from birth of the cellto division.

Transcription of genes encoding cell division proteins isnot strongly influenced by growth rate

The promoter–gfp fusions at amyE allow promoter activityto be monitored by the production of intracellular GFP.The reporter strains were cultured in three defined mediawith equal incubation temperature and aeration, and thedifferences in media composition resulted in three differ-ent doubling times (K) [K = 0.5, 1.0 and 2.0 massdoublings per hour (md h−1)]. Culturing in LB provided fora fast growth rate (K = 2.0 md h−1), culturing in adaptedcompetence media provided for the medium and slowgrowth rates (K = 1.0 and 0.5 respectively) (see Experi-mental procedures: strains, media and growth condi-tions). Cells were sampled at mid-exponential phase andproduction of GFP was determined by flow cytometry forgfp fusions to the promoters of ftsZ, ftsA, zapA, ftsW, noc,ezrA, minCD, dnaA and rrnB. Promoter activities werenormalized at a growth rate of 0.5 md h−1. The activity ofthe rrnB promoter clearly increased with growth rate, asreported earlier (Krasny and Gourse, 2004) (Fig. 2A). Theactivity of promoters for genes that are directly involved incell division (ftsZ, ftsA, zapA, ftsW) was found to bealmost equal at all growth rates. The activity of the pro-moter for minC (which is encoded in an operon withmreBCD and minD (Lee and Price, 1993), slightlyincreased with growth rate, as did the activity of the dnaApromoter. However, this increase in promoter activity wasnever more than the twice the activity measured at the lowgrowth rate of 0.5 md h−1.

Stable integration of the promoter–gfp constructs atamyE allows the study of strains over long time intervalsin the absence of antibiotics, but includes the risk that theinfluence of nearby genetic elements on the promoteractivity is missed. To verify the results from the amyEintegration strains, a subset of promoter–gfp fusions wascreated at the original locus. In addition, we extended ournative locus promoter library with promoters of genesencoding the regulatory CDP DivIVA and the late CDPFtsL. These fusions were created using the BaSysBioII

Fig. 1. Schematic representations of a few transcriptional fusionsto gfp.A. The entire promoter region of the ftsAZ operon + the ftsA geneincluding a stop codon was taken as the ftsZ promoter and placedbefore gfp followed by a double terminator. This construct wasintegrated at the amyE region by a double-cross-over event.B. The promoter region of the ftsAZ operon was taken as the ftsApromoter and placed before gfp followed by a double terminator.Also, this construct was integrated at the amyE region by adouble-cross-over event.C. The promoter region of the ftsAZ operon was taken as the ftsApromoter and placed before gfp followed by a single terminator. Bya single-cross-over event this construct was integrated as a copynext to the original gene.

Transcription of cell division genes in B. subtilis 3


vector that integrates as a copy next to the original locusin the B. subtilis chromosome by a single-cross-overevent (Botella et al., 2010). The resulting strains weregrown identically to the amyE integration strains and theactivity of the promoters for ftsA, ftsW, ftsL, divIVA andrrnB were determined by flow cytometry. This selectedgroup is representative for the cell division machinerysince it is comprised of genes that encode early-, mid- andlate-stage CDPs. Again, activity from the rrnB promoterincreased with growth rate, whereas the activity of thepromoters for cell division genes hardly varied (Fig. 2B).This result confirms that the measurements from theamyE locus provide a valid indication of the activities ofpromoters for cell division genes.

The forward scatter value, as measured during flowcytometry, provides an indication of the size of the cells inthe population. As expected, cell size increased withhigher growth rates (not shown) (Weart et al., 2007). It isknown that the diameter of B. subtilis cells remains

constant at different growth rates (Sharpe et al., 1998).Our data on the transcriptional activity of promoters forcell division genes suggests that B. subtilis needs a stableamount of CDPs at all growth rates. Indeed, equalamounts of FtsZ have been reported in both B. subtilisand E. coli cultured at different growth rates (Weart andLevin, 2003). The transcription of genes encoding pro-teins that are directly involved in division does not have toincrease with growth rate, because the diameter of thecell, across which division takes place, remains constantwhile cell length increases. In contrast, regulatory proteinsfor cell division can be required in larger numbers when acell is growing faster because cells are longer. To restrictthe position of Z-ring formation at mid-cell, more MinC andMinD may be necessary, since the area in which Z-ringformation needs to be blocked is larger. In summary, thedata from the promoter–gfp fusions indicate that the tran-scription of cell division genes in a bacterial population isconstant over different growth rates. Transcription of the

Table 1. Relevant plasmids and bacterial strains used in this study.

Strain/plasmid Relevant genotypea Reference

B. subtilis168 Trp−, 1A700 variant Veening et al. (2006)JWV133 168, amyE::PezrA–gfp, cm Gamba et al. (2009)A::PftsZ–gfp 168, amyE::PftsZ–gfp, cm This studyA::PftsA–gfp 168, amyE::PftsA–gfp, cm This studyA::PzapA–gfp 168, amyE::PzapA–gfp, cm This studyA::PftsW–gfp 168, amyE::PftsW–gfp, cm This studyA::PnoC–gfp 168, amyE::Pnoc–gfp, cm This studyA::PminC–gfp 168, amyE::PminC–gfp, cm This studyA::PdnaA–gfp 168, amyE::PdnaA–gfp, cm This studyA::PrrnB16s–gfp 168, amyE::PrrnB16s–gfp, cm This studyPftsA:PftsA–gfp 168, PftsA:PftsA–gfp, spec This studyPdivIVA:PdivIVA–gfp 168, PdivIVA:PdivIVA–gfp, spec This studyPftsL:PftsL–gfp 168, PftsL:PftsL–gfp, spec This studyPftsW:PftsW–gfp 168, PftsW:PftsW–gfp, spec This studyPrrnB16s:PrrnB–gfp 168, PrrnB16s:PrrnB16s–gfp, spec This study

Escherichia coliMC1000 F−, araD139 (ara ABC-leu)7679, galU, galK, lacX74, rspL, thi,BL21 F−, dcm, ompT, hsdS(rB− mB−), gal, [malB+]K-12(λS) StratageneTG1 Botella et al. (2010)

PlasmidspJWV017 spec, amyE, gfp+ (F64L, S65T, F99S, M153T, V163A), cat Veening et al. (2009)pGFP-ftsZ spec, amyE, PftsZ–gfp, cat This studypGFP-ftsA spec, amyE, PftsA–gfp, cat This studypGFP-zapA spec, amyE, PzapA–gfp, cat This studypGFP-ftsW spec, amyE, PftsW–gfp, cat This studypGFP-noc spec, amyE, Pnoc–gfp, cat This studypGFP-minC spec, amyE, PminC–gfp, cat This studypGFP-dnaA spec, amyE, PdnaA–gfp, cat This studypGFP-rrnB spec, amyE, PrrnB–gfp Veening et al. (2009)pBaSysBioII spec, LIC transcriptional fusion vector Botella et al. (2010)GFP-ftsA spec, PftsA, PftsA–gfp This studyGFP-divIVA spec, PdivIVA, PdivIVA–gfp This studyGFP-ftsW spec, PftsW, PftsW–gfp This studyGFP-ftsL spec, PftsL, PftsL–gfp This studyGFP-rrnB16s spec, PrrnB16s, PrrnB16s–gfp This study

a. cm, chloramphenicol resistance marker; tet, tetracycline resistance marker; amp, ampicillin resistance marker; km, kanamycin resistancemarker; spec, spectinomycin resistance marker; IPTG, isopropyl thio-β-D-galactoside.

4 E. N. Trip et al.


mreB-mreC-mreD-minC-minD operon, which containsthe genes coding for the MinC/D division site selectionsystem, appears slightly upregulated at higher growthrates. This was also observed for dnaA.

Transcription of cell division genes monitored by qrtPCR

To verify the results from the promoter fusion experimentsthe transcription of several cell division genes was moni-tored by an independent method: qrtPCR. Bacillus subtilis168 1A700 was grown at three different growth rates,exactly as described for the transcriptional fusion experi-ments, and at mid-exponential growth phase, total RNAwas isolated. The RNA was used to generate complemen-tary DNA (cDNA) to serve as a template for the qrtPCR.Equal amounts of cDNA were used as input material for

the qrtPCR. The ratios of cell division gene transcripts(ftsZ, ftsA, zapA, ftsW, minC, dnaA, divIVA, ezrA and noc)relative to their transcript level at the slowest growth ratewere calculated, as well as the transcript level comparedwith a housekeeping gene (veg) at the different growthrates. The housekeeping gene veg was selected as thisgene is constitutively transcribed during vegetative growth(Krasny and Gourse, 2004). As seen with the previousanalysis, the amount of rrnB transcripts increases stronglywith an increase in growth rate (Fig. 3A). Again theexpression level of minC was found to slightly increase,but not as strongly as rrnB16s (Fig. 3A). The other testedcell division genes displayed hardly any variation withgrowth rate (Fig. 3A), similar to the results obtained withthe transcriptional fusions. Comparing the amount of tran-scripts relative to those of the housekeeper gene veg thatis strongly and constitutively transcribed in cells growingin a variety of media (Fukushima et al., 2003; Nicolaset al., 2012), ftsA and ftsZ were found to be almost equaland most abundant (Fig. 3B). Indeed, FtsZ and FtsA

Fig. 2. A. Normalized expression of gfp transcriptional fusionsunder the control of regulatory regions of various cell divisiongenes and of rrnB16s (+) at different growth rates (slow ≈ 0.5 mdh−1, medium ≈ 1.0 md h−1, fast ≈ 2.0 md h−1). The reports come fromtranscriptional fusions to gfp integrated at the amyE locus. TotalGFP fluorescence per cell was taken as a measure fortranscription. By flow cytometry, 50.000 cells were measured;resulting in a normal distribution of which the peak-values weretaken to calculate transcription as described in the Experimentalprocedures. The transcription values for the slow growth rate areset to 1.B. Normalized expression of gfp transcriptional fusions under thecontrol of regulatory regions of various cell division genes and ofrrnB16s (+) at different growth rates (slow ≈ 0.5 md h−1,medium ≈ 1.0 md h−1, fast ≈ 2.0 md h−1). The reports come fromtranscriptional fusions to gfp integrated as a copy next to theoriginal gene. Data were collected as in (A).

Fig. 3. Amounts of mRNA transcripts as determined by qrtPCR ofvarious cell division genes and rrnB16s (+).A. Transcript levels were normalized to the level of transcription ofthe same gene at the lowest growth rate.B. Transcript levels were normalized to the level of thehousekeeping gene veg at different growth rates (slow≈ 0.5 md h−1, medium ≈ 1.0 md h−1, fast ≈ 2.0 md h−1).



protein copy numbers are relatively high in B. subtilis,estimated at 5000 and 1000 respectively (Feucht et al.,2001). Also, relatively large amounts of transcripts werefound for minC. As MinC binds to FtsZ in a 1:1 ratio(Okuno et al., 2009) and MinC has to cover both poles ofthe cell, a similar number of MinC and FtsZ transcriptsmakes sense. Fewer transcripts were found for dnaA,divIVA, ezrA and noc and very little zapA and ftsW tran-scripts were present. It appears likely that either the pro-teins that are encoded by these genes are only required insmall numbers or the mRNAs are very stable and/or effi-ciently translated.

Cell cycle-dependent increase in promoter activityensures constant CDP concentrations

The results above show that transcription of the genescoding for CDPs is not significantly affected by the growthrate (Figs 1–3). However, these experiments do notexclude the possibility that there is cell cycle-dependenttranscription. For example, using synchronized cultures, it

was suggested that transcription of ftsZ of E. coli oscil-lates during the cell cycle by two proximal promoters(Garrido et al., 1993) and it is well documented thatC. crescentus ftsZ is transcribed in a cell cycle-dependentmanner (Sackett et al., 1998; Quardokus et al., 2001). Toexamine whether transcription of ftsZ shows cell cycle-dependent oscillations, we turned to automated live celltime-lapse fluorescence microscopy. With this technique,cells are grown directly under a microscope and phasecontrast and GFP fluorescence images are taken at inter-vals and movies can be generated and analysed (de Jonget al., 2011). As shown in Fig. 4A and Movie S1, a fewcells grow out and divide, and within 12 h form smallmonolayered microcolonies consisting of a couple ofhundred cells. In the case of the PftsZ–gfp reporter strain,a clear increase in average population fluorescence canbe observed after approximately 600 min of growth, coin-ciding with the onset of sporulation (Fig. 4B). As PftsZcontains a second, sporulation-dependent, promoter(Gonzy-Treboul et al., 1992), this further highlights thestrength of this technique.

Fig. 4. Time-lapse fluorescence microscopyto assess CDP gene expression as a functionof the cell cycle.A. Stills of a growing microcolony of B. subtiliscells harbouring the PftsZ–gfp transcriptionalfusion. GFP fluorescence remains constantduring vegetative growth, but increasessubstantially upon entry into sporulation. Viacomputational analysis, data from individualcells in the growing microcolony such asfluorescence intensity and cell length andbirth times (as a measure for cell-cycleprogression) can be extracted.B. A graphical representation of the time-lapsefluorescence microscopy images in the upperpanel. The average GFP fluorescence of allthe cells in the microcolony is displayedagainst time. Upon entry into sporulation, anincrease in ftsZ transcription is clearlydiscernible.

6 E. N. Trip et al.


To assess whether transcription of ftsZ shows any cellcycle-dependent expression pattern, we first had toexamine whether the B. subtilis cell-cycle progresses lin-early or exponentially in time relative to the cell length. Todo so, we collected more than 900 data points of expo-nentially growing B. subtilis cells and plotted cell length asa function of the cell-cycle progression in time (seeExperimental procedures). Interestingly, the data pointsare much a better fit to an exponential curve than to a line(Fig. S1; R2 = 0.997 versus 0.982 respectively), indicatingthat, as cells progress in the cell cycle, they more rapidlyincrease in cell length. Interestingly, this also holds true forE. coli (Wang et al., 2010) suggesting that rod-shapedcells that can also elongate along the periphery of the cellhave an exponential mode of cell length growth as wasshown previously using more invasive techniques thatrequire cell fixation such as transmission electron micros-copy (Hariharan et al., 1982; Burdett et al., 1986;Kirkwood and Burdett, 1988; Koch, 1993). These resultsindicate that cell-cycle progression in B. subtilis can bestbe plotted as a function of cell length (Fig. S1).

Having established a robust way to categorize cells intotheir correct cell-cycle progression state (Fig. S1 andExperimental procedures), we collected data from threeindependent movies of the PftsZ–gfp strain. Only expo-nentially growing cells were taken into account (generallycells between approximately 250 and 500 min after thestart of the experiment). Plotting all data points as celllength versus cell-cycle progression (as a function of celllength) showed a straight line, both for the B. subtilisPftsZ–gfp data set and for a large E. coli data set for aninducible, non-cell-cycle-regulated promoter (Stewartet al., 2005), again verifying that our mathematical routineto analyse single cells from movies is robust (Fig. S2Aand B). Plotting the average GFP fluorescence (mean ofthe fluorescence level per pixel) from the ftsZ promoter asa function of cell-cycle progression showed a straight line(Fig. 5A). This indicates that the concentration of GFPprotein, and by inference FtsZ protein, does not fluctuateduring the cell cycle. Doing the same analysis on a movieof a late cell division reporter, PftsW–gfp, showed thesame result indicating the absence of a cell cycle-dependent manner of transcriptional regulation (Fig. 5C).We have shown previously that the sda gene is cell cycleregulated via transcriptional activation of DnaA (Veeninget al., 2009). Analysing a movie of a strain carrying aPsda–GFP fusion showed that the concentration of GFPfluctuates as a function of the cell cycle, as expected(Fig. 5E).

While average GFP levels are a good indication for theconcentration of the protein, it does not provide directinformation for the promoter activity (Rosenfeld et al.,2005; 2006; Zaslaver et al., 2009). Therefore, we plottedthe normalized promoter activity of PftsZ–gfp as a func-

tion of cell-cycle progression. As shown in Fig. 5C, ftsZpromoter activity gradually increases with increasing cell-cycle progression. At the end of the cell cycle, ftsZ pro-moter activity is roughly twofold higher than at thebeginning of the cell cycle, while this is not the case for thecell cycle-dependent sda promoter which is already highat the beginning of the cell cycle (Fig. 5F). The trend ofhigher promoter activity at the end of the cell cycle alsoholds true for the ftsW promoter but to a lesser extent thanfor PftsZ (Fig. 5D). This might be because less data pointswere available for this strain compared with the PftsZstrain.

To examine whether increased promoter activity is ageneral phenomenon for constitutive promoters, we ana-lysed a large E. coli data set of cells harbouring a yfpfusion driven by the IPTG inducible PLlacO1 promoter fullyinduced with 1 mM IPTG (Stewart et al., 2005) using ourimage analysis algorithms. As shown in Fig. S2C, thisE. coli promoter is solely controlled by inducer level, andthe concentration of YFP is constant regardless of thecell-cycle state. Importantly, PLlacO1 promoter activity isalso approximately twofold higher at the end of the cellcycle than at cell birth (Fig. S2D).

Besides the presence of a greater number of RNApolymerases and ribosomes within a cell at the end of itscell cycle, a likely explanation for the enhanced promoteractivity is the increase in chromosome copy number, assuggested previously (Elowitz et al., 2002). After DNA-replication but before cell division, cells at the end of thecell cycle contain twice as many promoters as at thebeginning of the cell cycle. Thus, in the absence of anycell-cycle control (i.e. in the case of constitutive promot-ers), promoter activity is twice as high at the end of the cellcycle compared with the beginning.

Concluding remarks

In this report, we show that the transcription of cell divisiongenes in B. subtilis is not regulated by growth rate both ona population and on a single cell level, and not cell cycleregulated on the single cell level. Stable transcription ofcell division genes independent of growth rate ensuresconstant translation and a fixed amount of CDPs per cell.The cell length of a rapidly growing B. subtilis cell can betwice that of a slower counterpart, but its diameter is fixed(Sharpe et al., 1998; Weart and Levin, 2003), so theamount of CDPs required to from a single septum doesnot have to change when the cell increases its size.

By balancing its numbers of CDPs, B. subtilis demon-strates efficient division; it produces the critical amountsof proteins needed for this process, not more. Studies inwhich mild overexpression of an inhibitor or absence of astabilizer of the divisome can be counteracted by mildoverexpression or a gain of function mutation in another



divisome component also point to a minimal amount ofeach CDP being produced (Dai and Lutkenhaus, 1992;Gueiros-Filho and Losick, 2002; Geissler et al., 2003;Kawai and Ogasawara, 2006). Constant transcription,

which we confirmed with quantitative rtPCR could be fol-lowed by constant translation. The transcriptional fusionstrains provide evidence for this as this technique meas-ures both transcription and translation of GFP. The levels

Fig. 5. A, C and E. Average GFP fluorescence plotted against cell cycle progression (based on cell length) for the PftsZ–gfp, PftsW–gfp andPsda–gfp transcriptional fusions.B, D and E. Promoter activity (arbitrary units per minute) plotted against cell cycle progression (based on cell length) for the PftsZ–gfp,PftsW–gfp and Psda–gfp transcriptional fusions.For details, see Fig. S2 and Experimental procedures.

8 E. N. Trip et al.


of GFP remain fairly constant over various growth rates inthe transcriptional fusion strains for genes encodingCDPs. Other studies also demonstrated equal levels ofthe main CDP, FtsZ, in E. coli and B. subtilis across a widerange of growth rates (Weart and Levin, 2003). Further-more, on a population level, within the transcriptomes ofB. subtilis cultured at a wide range of growth-conditions,the amounts of transcripts for cell division genes remainconstant. Medium composition, temperature and stresshardly influence their quantities, highlighting the impor-tance of a balanced transcription for these genes thatencode for proteins with an essential function (Nicolaset al., 2012).

Since the levels of mRNA and CDPs remain unchangedat various growth rates, other factors must dictate thepace of division. A richer medium allows for faster growthand rapid metabolism. Substrate and energy for buildingthe septum may be more abundant and more readilyavailable. In B. subtilis the metabolic sensor UgtP couplescarbon availability to cell division by inhibiting the forma-tion of the Z-ring until the cell reaches a critical mass(Weart et al., 2007). Such a mechanism, combined withconcentration gradients of division inhibitors such as Minand Noc, would indeed be sufficient to activate a pool ofCDPs already present, to perform a round of division.

Interestingly, the transcription of genes encoding thecell division regulatory proteins MinCD is increasingslightly with growth rate, as measured by levels of GFP(Fig. 2A). Because of the increase in cell length,B. subtilis could need a higher amount of negative regu-lators to prevent the formation of a division site at thewrong place. Negative regulators are also required toprevent the dilution of the pool of FtsZ so that functionalrings can be formed at every potential division site –Z-ring frequency is greatly decreased in a min/noc doubleknockout (Wu and Errington, 2004; Rodrigues and Harry,2012). Also, MinC prevents formation of a second Z-ringright after division at the new cell pole (Gregory et al.,2008). With new-born cells being significantly larger at afast growth rate there may be a need for a higher concen-tration of MinC to counteract the formation of Z-ringsimmediately after division. In addition, the expression ofminCD is regulated by the same promoter as that formreBCD, genes that play a role in cell wall synthesisduring elongation (Typas et al., 2011). Upregulation ofmreBCD could be essential for the increase in cell lengthat high growth rates. The molecular mechanism drivingthe potential growth rate-dependent transcription of themreBCD-minCD operon remains elusive.

Our single cell experiments confirmed that ftsZ expres-sion is upregulated during sporulation and shows thatregulatory events can be identified with the methods weused. The finding that the transcription of cell divisiongenes in B. subtilis is balanced both at different growth

rates at a population level and throughout the cell cycle onthe single cell level, resulting in constant quantities of celldivision protein, strongly suggests cell division is regu-lated by other factors that do fluctuate and that can betterpace division like the metabolic state and substrate avail-ability. Because of their stable quantities, CDPs canrespond more rapidly and in a tightly organized fashion tothese fluctuating factors, resulting in intricate interplayand specific activity that determines the cell-cycleprogression.

Experimental procedures

Strains, media and growth conditions

Bacterial strains are listed in Table 1. Bacillus subtilis 1681A700 cells were grown with various doubling times inadapted Bacillus competence medium (CM) for slow (K = 0.5md h−1) and medium growth rates (K = 1.0 md h−1) and inLuria–Bertani broth (LB) for a fast growth rate(K = 2.0 md h−1). Adapted CM for slow growth is Spizizen’ssalt medium (SSM) containing glucose (0.1% w/v), trypto-phan (0.2 mg ml−1), MgSO4 (3 mM) and FeNH4 citrate(5.5∗10−5% w/v) (Spizizen, 1958). Adapted CM for mediumgrowth is SSM containing glucose (1.0% w/v), tryptophan(0.4 mg ml−1), casamino acids (0.2% w/v), MgSO4 (3 mM)and FeNH4 citrate (11.0∗10−5% w/v). When required,spectinomycin (50 μg ml−1) and chloramphenicol (5 μg ml−1)was added. During culture for analysis, no antibiotics wereadded to ensure normal ribosomal activity. Overnight cultureswere diluted to an optical density of 0.02 at 600 nm (OD600) inthe adapted competence medium or to 0.05 in LB. The cellswere incubated at 37°C with shaking and grown until mid-exponential growth phase (OD600 ≈ 0.15–0.4) for sampling.

Construction of amyE integration plasmids

To construct plasmid pGFP_ssrA-ftsA, a PCR using primersftsA-F + EcoRI and ftsA-R + NheI was performed, usingchromosomal DNA of B. subtilis 168 as a template. Theamplified fragment was subsequently cleaved with EcoRI andNheI and ligated into the corresponding sites of pGFP_ssrA(Veening et al., 2009), resulting in plasmid pGFP_ssrA-ftsA.To construct plasmid pGFP-ftsA, carrying the B. subtilis ftsApromoter region and a strong RBS fused with gfp+ (F64L,S65T, F99S, M153T, V163A (Scholz et al., 2000; Zellmeieret al., 2006), flanked by amyE regions, a PCR with theprimers gfp-R + FseI and ter-F + FseI was performed, usingplasmid pGFP_ssrA-ftsA as a template. Next the templatewas digested with DpnI and the amplified fragment was sub-sequently cleaved with FseI and self-ligated resulting inplasmid pGFP-ftsA.

To construct plasmid pGFP_ssrA-ftsZ, a PCR usingprimers ftsA_stop-F + NheI and ftsA_stop-R + NheI was per-formed, using chromosomal DNA of B. subtilis 168 as a tem-plate. The amplified fragment was subsequently cleaved withNheI and ligated into the corresponding site of pGFP_ssrA-ftsA, resulting in plasmid pGFP_ssrA-ftsZ. Note that theforward primer introduces a stop codon directly after the



ATG-start codon of ftsA and introduces a frameshift ensuringthat there is only a single functional copy of ftsA present in thegenome, when integrated at amyE. To construct plasmidpGFP-ftsZ, carrying the B. subtilis ftsA promoter region, anon-functional ftsA gene, and the intergenic region betweenftsA and ftsZ (together here called PftsZ), fused to gfp+,flanked by amyE regions, a PCR with the primers gfp-R + FseI and ter-F + FseI was performed, using plasmidpGFP_ssrA-ftsZ as a template. Next the template wasdigested with DpnI and the amplified fragment was subse-quently cleaved with FseI and self-ligated resulting in plasmidpGFP-ftsZ.

To construct plasmid pGFP_ssrA-zapA, a PCR usingprimers zapA-F + EcoRI and zapA-R + NheI was performed,using chromosomal DNA of B. subtilis 168 as a template. Theamplified fragment was subsequently cleaved with EcoRI andNheI and ligated into the corresponding sites of pGFP_ssrA,resulting in plasmid pGFP_ssrA-zapA. To construct plasmidpGFP-zapA, carrying the B. subtilis zapA promoter regionand a strong RBS fused with gfp+, flanked by amyE regions,a PCR with the primers gfp-R + FseI and ter-F + FseI wasperformed, using plasmid pGFP_ssrA-zapA as a template.Next the template was digested with DpnI and the amplifiedfragment was subsequently cleaved with FseI and self-ligated resulting in plasmid pGFP-zapA.

To construct plasmid pGFP-ftsW, carrying the B. subtilisftsW promoter region and a strong RBS fused with gfp+, aPCR with the primers ftsW-F + EcoRI and ftsW-R + NheI wasperformed, using chromosomal DNA of B. subtilis 168 as atemplate. The amplified fragment was subsequently cleavedwith EcoRI and NheI and ligated into the corresponding sitesof pJWV017 (Veening et al., 2009), resulting in plasmidpGFP-ftsW.

To construct plasmid pGFP_ssrA-noc, a PCR usingprimers noc-F + EcoRI and noc-R + NheI was performed,using chromosomal DNA of B. subtilis 168 as a template. Theamplified fragment was subsequently cleaved with EcoRI andNheI and ligated into the corresponding sites of pGFP_ssrA,resulting in plasmid pGFP_ssrA-noc. To construct plasmidpGFP-noc, carrying the B. subtilis noc promoter region and astrong RBS fused with gfp+, flanked by amyE regions, a PCRwith the primers gfp-R + FseI and ter-F + FseI was per-formed, using plasmid pGFP_ssrA-noc as a template. Nextthe template was digested with DpnI and the amplified frag-ment was subsequently cleaved with FseI and self-ligatedresulting in plasmid pGFP-noc.

To construct plasmid pGFP_ssrA-minC, a PCR usingprimers minC-F + NotI and minC-R + NheI was performed,using chromosomal DNA of B. subtilis 168 as a template. Theamplified fragment was subsequently cleaved with NotI andNheI and ligated into the corresponding sites of pGFP_ssrA,resulting in plasmid pGFP_ssrA-minC. To construct plasmidpGFP-minC, carrying the B. subtilis minCD promoter regionand a strong RBS fused with gfp+, flanked by amyE regions,a PCR with the primers gfp-R + FseI and ter-F + FseI wasperformed, using plasmid pGFP_ssrA-minC as a template.Next the template was digested with DpnI and the amplifiedfragment was subsequently cleaved with FseI and self-ligated resulting in plasmid pGFP-minC.

To construct plasmid pGFP_ssrA-dnaA, a PCR usingprimers dnaA-F + EcoRI and dnaA-R + NheI was performed

was performed, using chromosomal DNA of B. subtilis 168 asa template. The amplified fragment was subsequentlycleaved with EcoRI and NheI and ligated into the correspond-ing sites of pGFP_ssrA, resulting in plasmid pGFP_ssrA-dnaA. To construct plasmid pGFP-dnaA, carrying theB. subtilis dnaA promoter region and a strong RBS fused withgfp+, flanked by amyE regions, a PCR with the primers gfp-R + FseI and ter-F + FseI was performed, using plasmidpGFP_ssrA-dnaA as a template. Next the template wasdigested with DpnI and the amplified fragment was subse-quently cleaved with FseI and self-ligated resulting in plasmidpGFP-dnaA.

All constructs were sequence verified.

Construction of amyE integrated B. subtilis strains

Bacillus subtilis strains A::P–gfp, A::PftsZ–gfp, A::PftsA–gfp,A::PzapA–gfp, A::PftsW–gfp, A::Pnoc–gfp, A::PminC–gfp,A::PdnaA–gfp and A::PrrnB–gfp were obtained by a double-cross-over recombination event between the amyE regionslocated on the pJWV017, pGFP-ftsZ, pGFP-ftsA, pGFP-zapA, pGFP-ftsW, pGFP-noc, pGFP-minC, pGFP-dnaA andpGFP-rrnB plasmids and the chromosomal amyE gene ofstrain 168 respectively (Veening et al., 2009). Transformantswere selected on LB agar plates containing chloramphenicol(Cm) after overnight incubation at 37°C. Correct integrationinto the amyE gene was tested and confirmed by lack ofamylase activity upon growth on plates containing 1% starch.

Construction of strains with a transcriptional fusionreporter at the original locus

The pBaSysBioII transcriptional fusion vector was used toconstruct the plasmids for transcriptional fusions that inte-grate as copies at the original locus (Botella et al., 2010). Thisplasmid has a SmaI restriction site, flanked by ligase inde-pendent cloning (LIC) sites, upstream of gfp. After restrictionby SmaI, promoter regions of interest can be cloned into thissite by LIC to control the transcription of gfp. The promoterregion of interest targets the integration of the pBaSysBioIItranscriptional fusion vector into the B. subtilis chromosomeas a copy next to the original promoter region by a single-cross-over event. The promoter regions of ftsA, divIVA, ftsW,yllB (ftsL operon) and rrnB16s were amplified by PCR, usingchromosomal DNA from B. subtilis 168 as a template, withthe appropriate primers listed in Table 2. By primer-design,LIC-sites were added at the ends of the fragments. Also,strong identical ribosome binding sites (RBS) were placedbetween the 3′ ends of the promoter-regions and the 3′LICsite. For LIC, 0.2 pmol of each fragment was incubated with2.5 mM dTTP and T4 DNA polymerase. The pBaSysBioIIplasmid was linearized using SmaI, gel-purified, and treatedwith T4 DNA polymerase and 2.5 mM dATP. A mix of 5 ngprepared plasmid with 15 ng of prepared fragment was usedto transform E. coli TG1, using standard techniques.Transformants were selected on LB plates containingspectinomycin (50 μg ml−1). Subsequently, these plasmidswere isolated from E. coli TG1 and correctness of the insertswas verified by sequencing. The correct plasmids were intro-duced to a non-chaining B. subtilis 168 strain (1A700)(Veening et al., 2006), using standard procedures

10 E. N. Trip et al.


(Anagnostopoulos and Spizizen, 1961) with selection forspectinomycin (50 μg ml−1). The integration of the vectors atthe desired locus (as a copy next to the original) was verifiedby PCR using primers designed for a location upstream of thepromoter region and for gfp on the vector (Table 2).

Flow cytometry

The various promoter-reporter strains were cultured at theconditions as described above; for each strain, cells were

obtained from cultures with slow (K = 0.5 md h−1), medium(K = 1.0 md h−1) and fast growth rate (K = 2.0 md h−1) at mid-exponential growth phase (OD600 ≈ 0.15–0.4). The cells werewashed and diluted in cold, sterile PBS and measured imme-diately. Flow cytometry was performed with a FACSCANTOflow cytometer (BD Biosciences). Parameters that weremeasured are forward scatter (FSC), side scatter (SSC),GFP fluorescence (FL-1) and the ratio FL-1/FSC, 50.000cells were measured. The flow cytometry data were analysedwith the WinMDI 2.9 software. From the resulting normal

Table 2. Oligonucleotides.a

Primer Sequence (5′ to 3′) Restriction or LIC site

ftsA-F + EcoRI GCGCGAATTCCAGACCACTGGACGTACAACAGC EcoRIftsA-R + NheI GCGCGCTAGCCATAGTAGTTCCTCCTTACACATTTCGATCATTTCTATTCTATTATTTGC NheIgfp-R + FseI GCTCAATGGCCGGCCTTATTATTTGTAGAGCTCATCCATGCCATG FseIter-F + FseI CGATAGGCCGGCCAATAAAGATCCTAGGACGCCGCCAAG FseIftsA_stop-F + NheI GCGCGCTAGCATGTAACAACAATGAACTTTACG NheIftsA_stop-R + NheI GCGCGCTAGCCATAGTAGTTCCTCCTTATCTGCCGAATGACTATCTATCTATTCC NheIzapA-F + EcoRI GCGCGAATTCCAGCAGGCTGTACGGAATGG EcoRIzapA-R + NheI GCGCGCTAGCCATAGTAGTTCCTCCTTACCGCGAGAATCCTAGTTCTTATC NheIftsW-F + EcoRI CGCAATTAAGGTCGGAATTCCTGC EcoRIftsW-R + NheI GCGCGCTAGCCATAGTAGTTCCTCCTTAGCTTCTAACTTCCATTATCTATGG NheInoc-F + EcoRI GCGCGAATTCGGTGAACACGATTTGTG EcoRInoc-R + NheI GCGCGCTAGCCATAGTAGTTCCTCCTTATTAAAAACTGCCATTTTTATAATTC NheIminC-F + NotI GCATGCGGCCGCTGTCAAGCATCCCGGGAATC NotIminC-R + NheI GCGCGCTAGCCATAGTAGTTCCTCCTTATTTCTTAAAGCAAAAATACC NheIdnaA-F + EcoRI GCGCGAATTCCGTATCCACAGAGGTTATCGACAAC EcoRIdnaA-R + NheI GCGCGCTAGCCATAGTAGTTCCTCCTTACCTTTTCTTAGAAAATGGCGTCTTTCTC NheIftsA-F + LIC-F CCGCGGGCTTTCCCAGCCAGACCACTGGACGTACAAC LIC-FftsA-R + LIC-R GTTCCTCCTTCCCACCCACATTTCGATCATTTCTATTCTATTATTTGC LIC-RdivIVA-F + LIC-F CCGCGGGCTTTCCCAGCCGTCCGTTATGCTGACAAGT LIC-FdivIVA-R + LIC-R GTTCCTCCTTCCCACCATTTTTACATTTCAGTACGGTTATCTAG LIC-RftsW-F + LIC-F CCGCGGGCTTTCCCAGCCGCAATTAAGGTCGGAATTC LIC-FftsW-R + LIC-R GTTCCTCCTTCCCACCAGCTTCTAACTTCCATTATCTATGG LIC-RyllB-F + LIC-F CCGCGGGCTTTCCCAGCCATTTTCTTTTCAAAATCAACATCAAG LIC-FyllB-R + LIC-R GTTCCTCCTTCCCACCAACTCACCACTATTCACCACTTC LIC-RrrnB16s-F + LIC-F CCGCGGGCTTTCCCAGCACATCCCTACGCCA LIC-FrrnB16s-R + LIC-R GTTCCTCCTTCCCACCGCTTTACTAATATAACATTCAGC LIC-RUpstream PftsA-F TAAAAGTGCTTGCCAAAGATUpstream PdivIVa-F CTGCTGTTACGGGACAUpstream PftsW-F CGGATCTGTATTCGCATTUpstream PyllB-F GGTGATGGGCTTCAAAAUpstream PrrnB16s-F GTTCCACGCACTGCgfpmut3-R TTCTCCTTTACGCATAGTAGftsA 250 bp fragment-F AACAACAATGAACTTTACGTCftsA-R GATATATTTCCATTAACGCCGftsZ-F TTGGAGTTCGAAACAAACATAftsZ-R TTCTTCAGCGGCTTTTTTzapA-F TCTGACGGCAAAAAAACAzapA-R TCCTTTTCTTTAAGCTGACGminC-F AAGACCAAAAAGCAGCAATAminC-R ATGAACAAATCTTTCTTTGAftsW-F TTAAAAAAAATGCTAAAATCTTATGATTAftsW-R AAGCAGTATCCCCTTCTGdivIVA-F CCATTAACGCCAAATGATATTCdivIVA-R ATTGCGTTTACGTCTTCArrnB16s-F AAGTCGAGCGGACAGGTGrrnB16s-R AGTGTGGCCGATCACnoc-F CCGGCAATTATTAAGGATTTTnoc-R GGTGATTTTCTTCTCCATAATCveg-F GCGAAGACGTTGTCCGATATTveg-R CTCTGAGCTTGCGGCGTCATC

a. Relevant restriction or LIC sites are underlined.



distribution (of 50.000 measurements) the peak medianvalues for total GFP per cell (Fl-1) were taken as a measurefor transcription. These values were normalized to theaverage value for the slow growing culture (K = 0.5 md h−1)set to one.

Quantitative real-time reverse-transcription PCR

RNA was isolated from B. subtilis 168 1A700wt. This strainwas cultured at the conditions as described above; cells wereobtained from cultures with slow (K = 0.5 md h−1), medium(K = 1.0 md h−1) and fast growth rate (K = 2.0 md h−1) at mid-exponential growth phase (OD600 ≈ 0.15–0.4). The cells weredisrupted with glass beads (Ø 0.25 mm) using the FP120FastPrep shaker (ThermoSavant). Total RNA was isolatedusing TRIZOL reagent (Chomczynski and Sacchi, 1987), andpurified using the TURBO DNA-free KIT (Ambion). cDNA wasprepared from 500 ng of this DNA-free RNA with the iSCRIPTcDNA synthesis kit (Bio-Rad). The cDNA was used as tem-plate for qrtPCR. For all qrtPCR reactions, equal amounts ofcDNA were used. Primers were designed to have identicalTms of 57°C and to multiply 250 bp fragments from ftsA, ftsZ,minC, ftsW, divIVA, ezrA, noc, dnaA, veg and rrnB16s.qrtPCRs were performed in triplo for each gene and everygrowth condition with the SensiMix SYBR Hi-ROX kit(Bioline) in a Mini Opticon real-time PCR system (Bio-Rad).The quantities of mRNA from specific cell division genes werecompared with the housekeeper veg per growth condition.These ratios were then compared between the differentgrowth rates.

Time-lapse fluorescence microscopy

Cells were grown at 30°C in Spizizen minimal medium andinoculated onto a thin semi-solid agarose matrix based on achemically defined medium and attached to a microscopeslide as described (Veening et al., 2008). Microscope slideswere incubated on a temperature controlled automatedDeltavision RT microscope (Applied Precision). Images wereobtained up to 24 h with a CoolSNAP HQ2 (Princeton Instru-ments) at 63× or 100× magnification. Fluorescent and phase-contrast images were recorded at specified intervals,basically as described (de Jong et al., 2011). To preventphototoxicity, the excitation light (480–500 nm for GFP) waslimited to 32% of the output of a 100 W Hg-vapour lamp byneutral density filters with an exposure time of 0.8 s. Emis-sion wavelengths were 509–547 nm (GFP) (filters fromChroma). Microscopy images were prepared for publicationusing ImageJ (http://rsb.info.nih.gov/ij/) and CorelDRAW X3(Corel Corporation).

Microscopy data analysis

Cells were automatically segmented and measured usingcustom BHV software (Stewart et al., 2005; de Jong et al.,2011) and information regarding cell length, width, fluores-cence intensity, orientation, geographical location within thecolony, and the complete lineage history of each cell wasrecorded. The software was supplemented for this paper with

new Python routines to calculate promoter activity and todefine cell-cycle progression. The Python code is availableupon request.

Promoter activity

To calculate the specific promoter activity for each reporter ateach cell-cycle progression category, we used an approachsimilar as pioneered by the Alon and Elowitz laboratories(Rosenfeld et al., 2005; 2006). In brief, to measure theaverage GFP production rate between movie frames (pro-moter activity), we took the difference in total GFP level per cellbetween consecutive time points and divided that by theirseparating time interval (8 min). This approach assumes thatthe half-life or photobleaching of GFP is insignificant within thetimescale of a cell’s life and that GFP is only diluted by cellgrowth. Western blot analysis has shown a half-life of 1h forGFP in B. subtilis while the in vivo half-life or photobleachingproperties of GFP under our imaging conditions wereneglectable (Veening et al., 2009). Thus, the abovementionedapproach is a valid one.

Cell-cycle progression

To determine the cell-cycle state (or cell-cycle progression) ofeach data point within a movie, only data from cells growingexponentially were taken into account for the analysis (typi-cally data of cells between 250 and 550 min after start of theexperiment, after cells have adapted and before cells entersporulation). Since the birth and division of each cell in thedata set is known, a cell’s cell-cycle progression can bedefined and binned in categories relative to the total time ittook this cell to complete one full cell cycle (from birth todivision, Fig. S1). For instance, the data of all newly borncells within the first 10% of their total cell-cycle progressiontime (the relative time it takes cells to divide again) are plottedin the 10% category in Fig. S1, and all cells between 10% and20% of their cell-cycle progression are plotted as the 20%category. Since B. subtilis shows an exponential increase incell length progression during its cell cycle, we plotted cell-cycle progression as a function of cell length (Fig. S1). Forevery GFP-reporter movie, the binned cell-cycle progressiondata set can be plotted against mean fluorescence (whichreflects CDP concentration), the total fluorescence (whichreflects total CDP levels) or the promoter activity (reflectingtranscription of CDPs).

Acknowledgements

We would like to thank Emma Denham for providing us withthe pBaSysBioII vector, the accessory E. coli TG-1 cloningstrain and protocols.

Work in the Errington lab was funded by GrantBB/G015902/1 from the UK Biotechnology and BiologicalSciences Research Council. Work in the Veening lab isfunded by a Veni-fellowship from the Netherlands Organisa-tion for Scientific Research (NWO-ALW). Work in theScheffers lab is funded by a Vidi career award from NWO-ALW.



http://rsb.info.nih.gov/ij/

References

Adams, D.W., and Errington, J. (2009) Bacterial cell division:assembly, maintenance and disassembly of the Z ring. NatRev Microbiol 7: 642–653.

Anagnostopoulos, C., and Spizizen, J. (1961) Requirementsfor transformation in Bacillus subtilis. J Bacteriol 81: 741–746.

Botella, E., Fogg, M., Jules, M., Piersma, S., Doherty, G.,Hansen, A., et al. (2010) pBaSysBioII: an integrativeplasmid generating gfp transcriptional fusions for high-throughput analysis of gene expression in Bacillus subtilis.Microbiology 156: 1600–1608.

Burdett, I.D., Kirkwood, T.B., and Whalley, J.B. (1986)Growth kinetics of individual Bacillus subtilis cells and cor-relation with nucleoid extension. J Bacteriol 167: 219–230.

Camberg, J.L., Hoskins, J.R., and Wickner, S. (2009) ClpXPprotease degrades the cytoskeletal protein, FtsZ, andmodulates FtsZ polymer dynamics. Proc Natl Acad SciUSA 106: 10614–10619.

Chien, A.C., Zareh, S.K., Wang, Y.M., and Levin, P.A. (2012)Changes in the oligomerization potential of the divisioninhibitor UgtP co-ordinate Bacillus subtilis cell size withnutrient availability. Mol Microbiol 86: 594–610.

Chomczynski, P., and Sacchi, N. (1987) Single-step methodof RNA isolation by acid guanidinium thiocyanate-phenol-chloroform extraction. Anal Biochem 162: 156–159.

Dai, K., and Lutkenhaus, J. (1992) The proper ratio of FtsZ toFtsA is required for cell division to occur in Escherichia coli.J Bacteriol 174: 6145–6151.

Dajkovic, A., Lan, G., Sun, S.X., Wirtz, D., and Lutkenhaus, J.(2008) MinC spatially controls bacterial cytokinesis byantagonizing the scaffolding function of FtsZ. Curr Biol 18:235–244.

Elowitz, M.B., Levine, A.J., Siggia, E.D., and Swain, P.S.(2002) Stochastic gene expression in a single cell. Science297: 1183–1186.

Errington, J., Daniel, R.A., and Scheffers, D.J. (2003)Cytokinesis in bacteria. Microbiol Mol Biol Rev 67: 52–65.

Feucht, A., Lucet, I., Yudkin, M.D., and Errington, J. (2001)Cytological and biochemical characterization of the FtsAcell division protein of Bacillus subtilis. Mol Microbiol 40:115–125.

Fukushima, T., Ishikawa, S., Yamamoto, H., Ogasawara, N.,and Sekiguchi, J. (2003) Transcriptional, functional andcytochemical analyses of the veg gene in Bacillus subtilis.J Biochem (Tokyo) 133: 475–483.

Gamba, P., Veening, J.W., Saunders, N.J., Hamoen, L.W.,and Daniel, R.A. (2009) Two-step assembly dynamics ofthe Bacillus subtilis divisome. J Bacteriol 191: 4186–4194.

Garrido, T., Sanchez, M., Palacios, P., Aldea, M., andVicente, M. (1993) Transcription of ftsZ oscillates duringthe cell cycle of Escherichia coli. EMBO J 12: 3957–3965.

Geissler, B., Elraheb, D., and Margolin, W. (2003) A gain-of-function mutation in ftsA bypasses the requirement for theessential cell division gene zipA in Escherichia coli. ProcNatl Acad Sci USA 100: 4197–4202.

Gonzy-Treboul, G., Karmazyn-Campelli, C., and Stragier, P.(1992) Developmental regulation of transcription of theBacillus subtilis ftsAZ operon. J Mol Biol 224: 967–979.

Gregory, J.A., Becker, E.C., and Pogliano, K. (2008) Bacillussubtilis MinC destabilizes FtsZ-rings at new cell poles andcontributes to the timing of cell division. Genes Dev 22:3475–3488.

Gueiros-Filho, F.J., and Losick, R. (2002) A widely conservedbacterial cell division protein that promotes assembly of thetubulin-like protein FtsZ. Genes Dev 16: 2544–2556.

Hariharan, I.K., Czolij, R., and Wake, R.G. (1982) Conforma-tion and segregation of nucleoids accompanying cell lengthextension after completion of a single round of DNA repli-cation in germinated and outgrowing Bacillus subtilisspores. J Bacteriol 150: 861–869.

de Jong, I.G., Beilharz, K., Kuipers, O., and Veening, J.W.(2011) Live cell imaging of Bacillus subtilis and Streptococ-cus pneumoniae using automated time-lapse microscopy.J Vis Exp 53: pii: 3145. doi:10.3791/3145.

Kawai, Y., and Ogasawara, N. (2006) Bacillus subtilis EzrAand FtsL synergistically regulate FtsZ ring dynamics duringcell division. Microbiology 152: 1129–1141.

Kelly, A.J., Sackett, M.J., Din, N., Quardokus, E., and Brun,Y.V. (1998) Cell cycle-dependent transcriptional andproteolytic regulation of FtsZ in Caulobacter. Genes Dev12: 880–893.

Kirkwood, T.B., and Burdett, I.D. (1988) Estimating thegrowth pattern of micro-organisms in distinct stages of thecell cycle. J Theor Biol 130: 255–273.

Koch, A.L. (1993) The growth kinetics of Bacillus subtilis.Antonie Van Leeuwenhoek 63: 45–53.

Krasny, L., and Gourse, R.L. (2004) An alternative strategyfor bacterial ribosome synthesis: Bacillus subtilis rRNAtranscription regulation. EMBO J 23: 4473–4483.

Lee, S., and Price, C.W. (1993) The minCD locus of Bacillussubtilis lacks the minE determinant that provides topologi-cal specificity to cell division. Mol Microbiol 7: 601–610.

Moriya, S., Rashid, R.A., Rodrigues, C.D., and Harry, E.J.(2010) Influence of the nucleoid and the early stages ofDNA replication on positioning the division site in Bacillussubtilis. Mol Microbiol 76: 634–647.

Nicolas, P., Mader, U., Dervyn, E., Rochat, T., Leduc, A.,Pigeonneau, N., et al. (2012) Condition-dependenttranscriptome reveals high-level regulatory architecture inBacillus subtilis. Science 335: 1103–1106.

Okuno, T., Ogoh, M., Tanina, H., Funasaki, N., and Kogure,K. (2009) Direct monitoring of interaction betweenEscherichia coli proteins, MinC and monomeric FtsZ, insolution. Biol Pharm Bull 32: 1473–1475.

Quardokus, E., Din, N., and Brun, Y.V. (1996) Cell cycleregulation and cell type-specific localization of the FtsZdivision initiation protein in Caulobacter. Proc Natl Acad SciUSA 93: 6314–6319.

Quardokus, E.M., Din, N., and Brun, Y.V. (2001) Cell cycleand positional constraints on FtsZ localization and the ini-tiation of cell division in Caulobacter crescentus. MolMicrobiol 39: 949–959.

Rodrigues, C.D., and Harry, E.J. (2012) The Min system andnucleoid occlusion are not required for identifying the divi-sion site in Bacillus subtilis but ensure its efficient utiliza-tion. PLoS Genet 8: e1002561.

Rosenfeld, N., Young, J.W., Alon, U., Swain, P.S., andElowitz, M.B. (2005) Gene regulation at the single-celllevel. Science 307: 1962–1965.



Rosenfeld, N., Perkins, T.J., Alon, U., Elowitz, M.B., andSwain, P.S. (2006) A fluctuation method to quantify in vivofluorescence data. Biophys J 91: 759–766.

Rueda, S., Vicente, M., and Mingorance, J. (2003) Concen-tration and assembly of the division ring proteins FtsZ,FtsA, and ZipA during the Escherichia coli cell cycle. JBacteriol 185: 3344–3351.

Sackett, M.J., Kelly, A.J., and Brun, Y.V. (1998) Orderedexpression of ftsQA and ftsZ during the Caulobactercrescentus cell cycle. Mol Microbiol 28: 421–434.

Sargent, M.G. (1975) Control of cell length in Bacillus subtilis.J Bacteriol 123: 7–19.

Scholz, O., Thiel, A., Hillen, W., and Niederweis, M. (2000)Quantitative analysis of gene expression with an improvedgreen fluorescent protein. p6. Eur J Biochem 267: 1565–1570.

Sharpe, M.E., Hauser, P.M., Sharpe, R.G., and Errington, J.(1998) Bacillus subtilis cell cycle as studied by fluores-cence microscopy: constancy of cell length at initiation ofDNA replication and evidence for active nucleoid partition-ing. J Bacteriol 180: 547–555.

Spizizen, J. (1958) Transformation of Biochemically DeficientStrains of Bacillus subtilis by Deoxyribonucleate. Proc NatlAcad Sci USA 44: 1072–1078.

Stewart, E.J., Madden, R., Paul, G., and Taddei, F. (2005)Aging and death in an organism that reproduces by mor-phologically symmetric division. PLoS Biol 3: e45.

Typas, A., Banzhaf, M., Gross, C.A., and Vollmer, W. (2011)From the regulation of peptidoglycan synthesis to bacterialgrowth and morphology. Nat Rev Microbiol 10: 123–136.

Veening, J.W., Kuipers, O.P., Brul, S., Hellingwerf, K.J., andKort, R. (2006) Effects of phosphorelay perturbations onarchitecture, sporulation, and spore resistance in biofilmsof Bacillus subtilis. J Bacteriol 188: 3099–3109.

Veening, J.W., Stewart, E.J., Berngruber, T.W., Taddei, F.,Kuipers, O.P., and Hamoen, L.W. (2008) Bet-hedging andepigenetic inheritance in bacterial cell development. ProcNatl Acad Sci USA 105: 4393–4398.

Veening, J.W., Murray, H., and Errington, J. (2009) A mecha-nism for cell cycle regulation of sporulation initiation inBacillus subtilis. Genes Dev 23: 1959–1970.

Wang, P., Robert, L., Pelletier, J., Dang, W.L., Taddei, F.,Wright, A., and Jun, S. (2010) Robust growth ofEscherichia coli. Curr Biol 20: 1099–1103.

Weart, R.B., and Levin, P.A. (2003) Growth rate-dependentregulation of medial FtsZ ring formation. J Bacteriol 185:2826–2834.

Weart, R.B., Nakano, S., Lane, B.E., Zuber, P., and Levin,P.A. (2005) The ClpX chaperone modulates assembly ofthe tubulin-like protein FtsZ. Mol Microbiol 57: 238–249.

Weart, R.B., Lee, A.H., Chien, A.C., Haeusser, D.P., Hill,N.S., and Levin, P.A. (2007) A metabolic sensor governingcell size in bacteria. Cell 130: 335–347.

Wu, L.J., and Errington, J. (2004) Coordination of cell divisionand chromosome segregation by a nucleoid occlusionprotein in Bacillus subtilis. Cell 117: 915–925.

Zaslaver, A., Kaplan, S., Bren, A., Jinich, A., Mayo, A., Dekel,E., et al. (2009) Invariant distribution of promoter activitiesin Escherichia coli. PLoS Comput Biol 5: e1000545.

Zellmeier, S., Schumann, W., and Wiegert, T. (2006) Involve-ment of Clp protease activity in modulating the Bacillussubtilis sigmaw stress response. Mol Microbiol 61: 1569–1582.

Supporting information

Additional Supporting Information may be found in the onlineversion of this article at the publisher’s web-site:

Fig. S1. Defining cell-cycle progression and exponentiallength growth.A. Cell-cycle progression defined as a function of time sincebirth and new cell division (left). Cell-cycle progressiondefined as a function of cell length (% of maximal cell lengthat time prior to new cell division) (right).B. More than 900 data points of exponentially growingB. subtilis cells are plotted as a function of the cell-cycleprogression in time (see Experimental procedures). The datapoints are much a better fit to an exponential curve than to aline (R2 = 0.997 versus 0.982 respectively),Fig. S2. Validation of our cell-cycle progression softwareroutines.A. All data points of B. subtilis PftsZ–gfp (∼ 900) were plottedas a function of cell-cycle progression based on cell lengthversus cell length. A clear straight line is visible demonstrat-ing that the binning procedure works as expected. Error barsrepresent standard error of the mean.B. Same as for (A) but now on a large (15593 data points)E. coli data set (Stewart et al., 2005).C. Average fluorescence from cells carrying an IPTG induc-ible YFP construct grown in the presence of 1 mM IPTG as afunction of cell-cycle progression. Same data set as used for(B). Error bars indicate standard error of the mean.D. Promoter activity (normalized GFP production per minute)as a function of cell-cycle progression for the E. coli data set.Error bars indicate standard error of the mean. Note theapproximate twofold increase in PA at the end of the cellcycle.Movie S1. This film shows approximately 20 h of the growthof a B. subtilis microcolony harbouring the PftsZ–gfp con-struct. Images were taken every 8 min. Scale bar: 10 μm.



Balanced transcription of cell division genes in Bacillus subtilis as revealed by single cell...

Documents

Transcript of Balanced transcription of cell division genes in Bacillus subtilis as revealed by single cell...